mechramc / Orion
Local AI runtime for training & running small LLMs directly on Apple Neural Engine (ANE). No CoreML. No Metal. Offline, on-device fine-tuning & inference on M-series silicon.
Why this job: says: fine tuning
What it does
Apple ships a dedicated Neural Engine (NPU) in every iPhone, iPad, and Mac sold since 2020 — over 2 billion devices. Each has a dedicated ML accelerator capable of ~19 TFLOPS fp16 (Apple claims 38 TOPS INT8, but maderix showed the ANE dequantizes to fp16 before computation). Yet Apple only exposes it through CoreML, which:
Get started
No single install command. Read the steps in the README (opens in new tab)
- Maintenance
- Active
- Last commit 2 months ago
- Stars
- 133
- On GitHub
- Licence
- MIT
- Allows business use
- Downloads for
- None listed
- No release files
- Latest version
- None yet
- No releases
- Built with
- Objective-C
- 84% of the code
Is it looked after?
- Contributors
- 1
- Last 90 days
- Issues
- 14% closed
- 6 still open
- Pull requests
- 1 merged
- 3 waiting
- Releases
- None
- Never published
- No contributing guide
- No security policy
- No code of conduct
More: languages, topics and project record
Languages
- Objective-C 84%
- C 12%
- Python 3%
- Makefile 1%
Topics
- ai
- apple-silicon
- llm-on-device
- machine-learning
- neural-engine
- reverse-engineering
Project record
- JobFine-tuning
- StatusListed
- Commits49
- Forks16
- Watching3
- Stars in 28 days+4
- CreatedMar 3, 2026
- First seen by usNov 7, 2026
- CheckedNov 7, 2026